JPH08298599A - Image encoding method and device therefor - Google Patents

Abstract

PURPOSE: To realize an image encoding method and device therefor by which high encoding efficiency can be obtained. CONSTITUTION: The picture element in the vicinity of the noticed picture element x of image data is defined as a specified picture element a and the differences (b-a), (c-a) of the density values of the peripheral picture elements b and c of the noticed picture element x and the specified picture element a are determined. The sizes of the differences (b-a), (c-a) and a prescribed value (Th1) are compared (S3). When either one of the differences (b-a), (c-a) is below the prescribed value (Th1), a Markob model encoding is performed for the density difference (x-a) based on the density difference (x-a) of the noticed picture element and the specified picture element and the differences (b-a), (c-a) (S4). When both of the differences (b-a), (c-a) are more than the prescribed value (Th1), a a predictive encoding is performed for the noticed picture element x based on the error of the density average (a+b)/2} of the specified picture element a and the peripheral picture element b and the density value of the noticed picture element x (S5).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image encoding method and apparatus for encoding image data by lossless encoding.

[0002]

2. Description of the Related Art Conventionally, Markov model coding method and predictive coding method are known as lossless coding methods for images.
Among them, the Markov model encoding models an image as an m-fold Markov information source in which the luminance value of the pixel of interest is determined by the state of m pixels (reference pixels) around the pixel of interest,
This is a method of encoding using a different entropy encoding method (Huffman code, arithmetic code, etc.) for each state of peripheral m pixels, and high encoding efficiency can be expected.

The predictive coding method is a method of predicting a pixel value of a pixel of interest from peripheral pixels of the pixel of interest and entropy coding the difference between the predicted pixel value and the actual pixel value.
This predictive coding is a method whose processing is relatively simple.

[0004]

In the above-described Markov model coding, it is necessary to generate a Markov table showing the appearance probability of the symbol for each state, and when applied to an image with a large number of gradations, The size of this Markov table is a problem. For example, in the case of performing n-pixel reference Markov model coding on image data of 256 gradations, the number of states is 256 to the n-th power. Here, in order to reduce the entropy, it is necessary to increase the number n of reference pixels, but when the number n of reference pixels is increased, the number of states increases geometrically, which is difficult to realize. Therefore, it is common to reduce the number of states by grouping the states with a small number of appearances into one state, or by grouping the states having similar appearance probability distributions of symbols, but this is complicated. However, there is a drawback in that various types of processing are required.

On the other hand, while predictive coding is easy to process,
There is a drawback that high encoding efficiency cannot be obtained.

The present invention has been made in view of the above-mentioned conventional example, and an object of the present invention is to provide an image coding method and an apparatus therefor which can obtain high coding efficiency.

Another object of the present invention is to provide an image coding method and apparatus for efficiently coding image data by using both Markov coding and predictive coding.

Another object of the present invention is to provide an image coding method and apparatus which can reduce the scale of a table for Markov model coding.

[0009]

In order to achieve the above object, the image coding method of the present invention comprises the following steps. That is, an image encoding method for encoding image data, wherein a pixel in the vicinity of a target pixel of the image data is set as a specific pixel, and a difference between density values of the peripheral pixel of the target pixel and the specific pixel is calculated, When the difference is less than or equal to a predetermined value, a step of Markov model coding the density difference based on the density difference between the pixel of interest and the specific pixel and the difference; and when the difference is equal to or more than the predetermined value, the specific pixel And a step of predictively encoding the pixel of interest based on the average density of the peripheral pixels and the density value of the pixel of interest.

In order to achieve the above object, the image coding apparatus of the present invention has the following configuration. That is, an image encoding device that encodes image data, and an extracting unit that extracts a pixel of interest of the image data, a pixel in the vicinity of the pixel of interest, and peripheral pixels of the pixel of interest as a specific pixel and a reference pixel, respectively. Calculation means for obtaining the difference in density value between the peripheral pixel and the specific pixel, comparison means for comparing the difference with a predetermined value, the density difference based on the density difference between the pixel of interest and the specific pixel and the difference A Markov model coding means for coding a Markov model, and a predictive coding means for predictively coding the pixel of interest based on the density average of the specific pixel and the peripheral pixel and the density value of the pixel of interest,
According to the comparison result by the comparison means, there is a selection means for coding the pixel of interest by using one of the Markov model coding means and the prediction coding means.

[0011]

In the above structure, the pixel of interest of the image data, the pixel in the vicinity of the pixel of interest, and the peripheral pixels of the pixel of interest are extracted as the specific pixel and the reference pixel, respectively, and the difference between the density values of the peripheral pixel and the specific pixel is extracted. And the difference is compared with a predetermined value. Depending on the comparison result, the density difference between the pixel of interest and the specific pixel and the difference based on the difference are Markov model coded, or the average density of the specific pixel and the peripheral pixel and the pixel of interest. Based on the density value, it is selected whether the target pixel is to be predictively coded, and the target pixel is operated to be coded.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A preferred embodiment of the present invention will now be described in detail with reference to the accompanying drawings. In the present embodiment, a 16-value monochrome image will be described as an image to be encoded.

[First Embodiment] FIG. 2 is a diagram showing the positional relationship of pixels in the present embodiment. In FIG. 2, x represents the position of a pixel of interest to be encoded, a represents a specific pixel, and b and c represent positions of reference pixels.

FIG. 1 is a flow chart showing the flow of processing in this embodiment. The operation of this embodiment will be described below with reference to FIG. 2 and this flow chart.

In FIG. 1, first, in step S1, a pixel value x of interest and a specific pixel value a of the image data to be encoded,
The reference pixel values b and c are extracted, and in step S2, the difference value (b between the reference pixel values b and c and the specific pixel value a (b
-A) and (c-a) are calculated. Next, in step S3, the magnitudes of the difference values (ba) and (c-a) are compared with the threshold value Th1. Then, the difference values (ba), (c
If any of the magnitudes of −a) is equal to or larger than the threshold value Th1, the process proceeds to step S5, (a + b).
The pixel of interest is predictively coded with / 2 as the prediction value. on the other hand,
When both the difference values (ba) and (ca) are smaller than the threshold value Th1 in step S3, the process proceeds to step S4, and the difference value (x-a) is Markov model encoded. When the process of step S4 or step S5 is completed, the process proceeds to step S6, and it is determined whether or not the pixel of interest is the last pixel of the image to be encoded. If not, the process returns to step S1.

FIG. 3 is a diagram showing an outline of the code table used in step S5 of FIG. 1, and FIG. 4 is a diagram showing an outline of the Markov table used in step S4.

FIG. 3 shows that, when the magnitudes of the difference values (ba) and (ca) exceed the threshold value Th1, the probability of occurrence of the difference value of the predicted value with (a + b) / 2 as the predicted value is shown. The case where code allocation is performed is shown.

FIG. 4 shows that when the magnitudes of the difference values (b-a) and (c-a) do not exceed the threshold value Th1, the difference value (b-
The case where code allocation is performed according to the appearance probability of the difference value (x-a) in each state divided by the combination of (a) and (c-a) is shown.

In these tables, the image to be encoded may be pre-scanned before the encoding process, statistics of the pixel values thereof may be taken, and a Huffman code may be created based on the statistics, or from several types of images in advance. Alternatively, a general-purpose table may be determined based on the statistics and used.

The image coding process here is performed for each pixel in raster scan order. First, the above step S1
Then, the values of the pixel of interest x, the pixel a immediately before it, and the reference pixels b and c are extracted. Next, in step S2, the difference values (ba) and (ca) between the reference pixels b and c and the pixel a are obtained. Next, in step S3, the difference value (b
-A) and (c-a) are examined, and both are set to the threshold T
If it is less than h1 (for example, 3), the process proceeds to step S4, and if at least one is greater than or equal to the threshold value, the process proceeds to step S5. In step S5, {(a + b) / 2} is used as a prediction value, and the prediction difference value {x- (a + b) / 2} is encoded and output according to the code table shown in FIG. 3, and step S6
Then, it is checked whether or not the pixel of interest is the last pixel of the image, and if it is the last pixel, the process is ended. If not, the process from step S1 is executed for the next pixel.

In step S4, the difference value (b-
The states are divided by a) and (c-a), and the difference value (x-a) is encoded using the Markov table of FIG. 4 and output. Then, in step S6, it is checked whether or not the pixel of interest is the last pixel of the image, and if it is the last pixel, the encoding process is ended. If it is not the last pixel, the process from step S1 is executed for the next pixel. To do.

The scale of the Markov table of this embodiment is (5 × 5 × 31), as is apparent from FIG. 4, and includes the scale of the code table (FIG. 3) required for predictive coding. However, the size of the Markov table in the case of performing the Markov model coding with reference to the three pixels of a, b, and c by the conventional method can be made extremely smaller than the size of the Markov table, that is, 16 4 Further, although the table size can be reduced in this way, it can be expected that the reduction in coding efficiency will be slight.

[Second Embodiment] FIG. 5 shows a second embodiment of the present invention.
In the present embodiment, it is assumed that the image to be encoded is a 16-value monochrome image and the positional relationship of pixels is the same as that in the first embodiment.

In FIG. 5, step S11 is a step of extracting from the image the value of the pixel of interest x and the pixels a and reference pixels b and c immediately before it, and step S12 is the reference pixels b and c.
Is a step of obtaining the difference values (b-a) and (c-a) between the pixel and the specific pixel a. In step S13, the difference value (b-
a) and (c-a) are examined and compared with the threshold value Th1, step S14 is the difference value (ba),
This is a step of predictively encoding the pixel value x of interest with {(a + b) / 2} as a prediction value when any of the magnitudes of (c−a) exceeds the threshold Th1. Step S15 is a step of comparing the magnitude of the difference value (x-a) with the threshold value Th2, and step S16 is the step of comparing the difference value (x-a) with the threshold value T.
This is a step of performing Markov model coding of the difference value (x-a) when h2 is not exceeded. The step S17 is a step of Markov model coding an OB code indicating that the difference value (x-a) is not included in the Markov table when the magnitude of the difference value (x-a) exceeds the threshold Th2. Is. Further, step S18 is a step of predictively encoding the pixel value x of interest with a being a prediction value, step S1.
Step 9 is a step of determining whether or not the pixel of interest a is the last pixel of the image.

FIG. 3 is a diagram showing an example of the code table used in step S14, FIG. 6 is a diagram showing an example of the Markov table used in steps S16 and S17, and FIG. 7 is an example of the code table used in step S18. FIG.

FIG. 3 shows that when the difference values (b-a) and (c-a) exceed the threshold value Th1, {(a + b) / 2}.
It shows a case where the code is assigned according to the appearance probability of the prediction difference value with the prediction value as.

FIG. 6 shows that when the magnitudes of the difference values (ba) and (c-a) do not exceed the threshold value Th1, the difference value (b-
The case where code allocation is performed according to the appearance probability of the difference value (x-a) in each state divided by the combination of (a) and (c-a) is shown.

Further, FIG. 7 shows the difference values (b-a), (c-
a) does not exceed the threshold Th1 and the difference value (x
When the magnitude of −a) exceeds the threshold Th2, the code assignment is performed according to the appearance probability of the difference value (x−a).

In the second embodiment, as in the case of the first embodiment described above, these tables are generated by prescanning the image before the encoding process to obtain statistics and assign the Huffman code. Alternatively, a general-purpose code table may be determined in advance by collecting statistics from several types of images and then used.

Next, referring to the flowchart of FIG.
The operation of the embodiment will be described.

The image encoding process is performed for each pixel in raster scan order. First, in step S11, the pixel of interest x
And the pixel a immediately before that and the reference pixel values b and c are extracted from the image to be encoded. Next, in step S12, the difference values (ba) and (ca) between the reference pixels b and c and the pixel a are obtained. Next, in step S13, the magnitudes of the difference values (b-a) and (c-a) obtained in step S12 are checked. If any of these is less than the threshold Th1 (for example, 3), go to step S15. If not, the process proceeds to step S14.

In step S14, {(a + b) / 2}
As the prediction value, the prediction difference value {x- (a + b) / 2} is encoded and output according to the code table of FIG. 3, and the process proceeds to step S19. In step S15, the difference value (x
-A) is compared with the threshold value Th2 to determine the threshold value Th2.
If it is less than (for example, 5), the process proceeds to step S16, and if not, the process proceeds to step S17.

At step S16, the difference value (ba),
The state is divided into states according to the value of (c−a), the difference value (x−a) is encoded and output using the Markov table shown in FIG. 6, and the process proceeds to step S19. -A), (c-
The state is divided according to the value of a), and the OB code indicating that the difference value (x−a) is outside the range of “−4” to “4” is encoded using the Markov table of FIG. Output.
Next, the procedure proceeds to step S18, and the prediction difference value (x-a) is calculated using the code table of FIG. 7 with the specific pixel value a as the prediction value.
Is encoded and output. When each process is completed in this way, the process proceeds to step S19 to check whether the pixel of interest is the last pixel of the image to be encoded, and if so, the encoding process ends, and if it is not the last pixel, the process returns to step S11. ,
The same process as described above is executed for the next pixel.

In the second embodiment, as is clear from FIG. 6, the scale of the Markov table is (5 × 5 × 10).
Therefore, even if the scale of the code table required for predictive coding is included, the scale of the Markov model in the case of performing the Markov model coding of three-pixel reference of a, b, and c by the conventional method, that is, 16 It can be made extremely small compared to the fourth power of.

[Third Embodiment] FIG. 8 is a flow chart showing the processing flow of the third embodiment of the present invention. This third
In the embodiment, the image to be encoded is a 16-value monochrome image, and the pixel positional relationship is as shown in FIG.

In FIG. 8, S21 is a step of extracting the values of the pixel of interest x and the reference pixels a, b, c from the image, S2.
2 is the difference between the reference pixels a, b and c (ba), (c-
a) and (b-c) are determined, and S23 is a step in which the magnitudes of the difference values (b-a), (c-a), and (b-c) are checked and compared with the threshold value Th3. S24 is
When all of the difference values (b−a), (c−a), and (b−c) exceed the threshold value Th3, this is a step of predictively encoding the pixel value x of interest with (a + b−c) as a prediction value. Also S
25 is a step of coding x with Markov model by referring to a, b, and c when any of the difference values (ba), (ca), and (bc) does not exceed the threshold Th3. Is.
Further, S26 is a step of determining whether the pixel of interest is the last pixel of the image.

FIG. 10 is a diagram showing an example of the code table referred to in step S24 of FIG. 8, and FIG. 11 is step S2.
5 is a diagram showing an example of a Markov table referred to in FIG.

FIG. 10 shows the difference values (ba), (c-).
The difference value (x−) from the predicted value (a + b−c) in the case where one of the values a) and (b−c) is less than or equal to the threshold Th3.
a-b + c) shows the case where code allocation is performed according to the appearance probability, and FIG. 11 shows difference values (b-a), (c-a),
When all the sizes of (bc) exceed the threshold value Th3,
In each state divided by the combination of a, b, and c, the case where the code is assigned according to the appearance probability of the pixel value x of interest is shown.

In the third embodiment, as in the other embodiments described above, these tables may be generated by prescanning the image before the encoding process, taking statistics, and assigning a Huffman code. However, a general-purpose code table may be determined in advance by obtaining statistics from several types of images and used.

Next, the operation of the third embodiment will be described with reference to the flowchart of FIG.

Also in this third embodiment, the image coding process is performed for each pixel in the raster scan order. First, in step S21, the values of the pixel of interest x and the reference pixels a, b, c are determined from the image to be coded. Take it out. Next, step S22
To the difference value (ba) between the reference pixels a, b, c,
The magnitudes of (c-a) and (bc) are checked, and if all the difference values are equal to or greater than the threshold Th3 (for example, 2), step S24.
If any one of them is less than the threshold Th3, the process proceeds to step S25. In step S24, the reference pixel a,
A prediction value (a + b-c) is created from b and c, and the difference value (x-a-b + c) between the pixel value of interest and the prediction value is encoded according to the code table of FIG. 10 and output.

On the other hand, in step S25, the reference pixel values a, b, and c are divided into respective states, and the pixel value x of interest is encoded and output using the Markov table of FIG. When the process of step S24 or S25 is completed in this way, the process proceeds to step S26 to determine whether the pixel of interest is the last pixel of the image to be encoded, and if so, the encoding process is terminated. If not the last step, step S21
Then, the above process is executed.

As described above, according to the third embodiment, the scale of the Markov table can be reduced to about half the 16th power of four as compared with the case of performing the Markov model coding with reference to 3 pixels in the conventional method.

FIG. 12 is a block diagram showing the functional arrangement of the image coding apparatus according to the present embodiment. It is shown that all the arrangements of the above-described embodiment are included, but connections of the respective parts are omitted as necessary. It is possible.

In FIG. 5, reference numeral 101 denotes an image input unit for inputting image data, and the image data input from the image input unit is input to a target pixel extracting unit 102, a specific pixel extracting unit 103, and a reference pixel 104, respectively x, the specific pixel a, and the reference pixels b and c are extracted. A calculation unit 105 calculates pixel values (density values) input from the pixel-of-interest extraction unit 102, the specific-pixel extraction unit 103, and the reference pixel 104, and calculates (ba), (ca), (b). -C), and further, the values of (a + b) / 2, (a + b-c), (x-a), etc. are calculated.

Reference numeral 106 denotes a comparison unit, which compares the threshold values (Th1, Th2, Th3) and the like stored in the threshold value memory 107 with the calculated value input from the calculation unit 105. The comparison process in the comparison unit 106 is performed in step S of FIG.
3, the steps S13 and S15 of FIG. 5, and further the step S23 of FIG.

The Markov model coding unit 111 or the predictive coding unit 108 is selectively driven according to the result of the comparison. Reference numeral 111 denotes a Markov table used for Markov coding, which is configured as shown in FIGS. 4, 6 and 11, for example. A code table 110 is used for predictive coding, and is configured as shown in FIGS. 3, 7, and 10, for example.

The predictive coding unit 108, for example, as shown in step S5 of FIG. 1 or step S14 of FIG.
And the densities of the reference pixel b and the specific pixel a are input, and the average value of the target pixel, the specific pixel, and the reference pixel {(a + b) /
2} is obtained, and the pixel of interest is predictively coded by referring to the table 110 based on this density difference.

On the other hand, the Markov model coding unit 109
Is, for example, as shown in step S4 of FIG.
According to the comparison result in 06, (ba) and (c-
Based on a) and the difference value between the pixel of interest x and the specific pixel a, encoding is performed by referring to the table 111 in FIG.

Similarly, the predictive coding unit 108 performs predictive coding as shown in step S18 of FIG. 5 or step S24 of FIG. 8, and the Markov model coding unit 109 uses the predictive coding of FIG.
As shown in steps S16 and S17 in step S25 and step S25 in FIG. 8, the pixel of interest is Markov model encoded. The coded data thus predictively coded or Markov model coded is stored in a memory (not shown),
Alternatively, it is output via an interface unit or the like.

The present invention is not limited to the above-described embodiments. For example, in each of the above-mentioned embodiments, the pixel a immediately before the pixel x of interest, the pixel b immediately above the pixel x, the upper left corner or the upper right corner, as peripheral pixels. However, the positions of these pixels are not limited to this, and any pixels that can be referred to may be used, and, for example, pixels located farther away may be referred to.

In the above embodiment, the case of static Markov model coding has been described. However, a step of updating the Markov table according to the appearing symbol is added and applied to the dynamic Markov model coding. Is also good. Further, the state division method at the time of Markov model encoding is not limited to the above-mentioned drawings, and any state division method may be used as long as it can correctly determine the state at the time of decoding.

The present invention may be applied to a system composed of a plurality of devices or an apparatus composed of a single device. The present invention can also be applied to the case where it is achieved by supplying a program for implementing the present invention to a system or an apparatus.

As described above, according to the present embodiment, the Markov model coding and the predictive coding are selected and performed according to the difference value between the pixel value of interest and the peripheral pixel values thereof, whereby the Markov model code is selected. Since it is possible to reduce the number of conversion states, it is possible to reduce the scale of the Markov table referred to when the Markov model is encoded, and it is possible to save memory.

[0055]

As described above, according to the present invention, there is an effect that a high coding efficiency can be obtained.

Further, according to the present invention, there is an effect that the image data can be efficiently coded by using the Markov coding and the predictive coding together.

Further, according to the present invention, there is an effect that the scale of the table for Markov model coding can be reduced.

[0058]

[Brief description of drawings]

FIG. 1 is a flowchart showing an encoding process according to a first embodiment of the present invention.

FIG. 2 is a diagram showing a positional relationship of pixels in the first embodiment.

FIG. 3 is a diagram showing an outline of a code table used in the first and second embodiments of the present invention.

FIG. 4 is a diagram showing an outline of a Markov table in the first embodiment.

FIG. 5 is a flowchart showing the flow of processing in the second embodiment of the present invention.

FIG. 6 is a diagram showing an example of a Markov table according to the second embodiment of the present invention.

FIG. 7 is a diagram showing an example of a code table in the second embodiment.

FIG. 8 is a flowchart showing a processing flow of a third embodiment of the present invention.

FIG. 9 is a diagram showing a pixel of interest and a reference pixel position in the third embodiment.

FIG. 10 is a diagram showing an example of a code table in the third embodiment.

FIG. 11 is a diagram showing an example of a Markov table in the third embodiment.

FIG. 12 is a block diagram showing a functional configuration of an image coding apparatus according to an embodiment of the present invention.

[Explanation of symbols]

 101 Image Input Unit 102 Target Pixel Extraction Unit 103 Specific Pixel Extraction Unit 104 Reference Pixel Extraction Unit 105 Calculation Unit 106 Comparison Unit 107 Threshold 108 Predictive Encoding Unit 109 Markov Model Encoding Unit 110 Predictive Encoding Table 111 Markov Table

Claims (10)

[Claims] 1. An image encoding method for encoding image data, wherein a pixel in the vicinity of a pixel of interest of image data is set as a specific pixel, and a difference between density values of peripheral pixels of the pixel of interest and the specific pixel is obtained. And, when the difference is less than or equal to a predetermined value, a step of Markov model encoding the density difference based on the density difference and the difference between the pixel of interest and the specific pixel, and when the difference is the predetermined value or more, A method of predicting and coding the pixel of interest based on a density average of the specific pixel and the peripheral pixel and a density value of the pixel of interest, the image coding method. 2. An image encoding method for encoding image data, wherein a pixel in the vicinity of a pixel of interest of image data is set as a specific pixel, and a difference between density values of peripheral pixels of the pixel of interest and the specific pixel is obtained. And a step of comparing the density difference between the pixel of interest and the specific pixel with a second predetermined value when the difference is less than or equal to a first predetermined value, and the density difference is less than or equal to the second predetermined value. Markov model coding the density difference, when the density difference is equal to or more than the second predetermined value, the Markov model coding based on the density difference, the pixel of interest using the specific pixel Predictive-encoding, and when the difference is equal to or greater than the first predetermined value, predictively-encoding the pixel of interest based on an average of the specific pixel and the peripheral pixel and a density value of the pixel of interest. Characterized by having Image encoding method. 3. An image encoding method for encoding image data, wherein a pixel in the vicinity of a pixel of interest of the image data is set as a specific pixel, and a difference between density values of the peripheral pixel of the pixel of interest and the specific pixel is obtained. And, when the difference is less than or equal to a predetermined value, a step of Markov model coding the target pixel based on the density value of the target pixel and the density values of the specific pixel and the peripheral pixel, and the density difference is the predetermined value. In the above case, the method includes the step of predictively encoding the pixel of interest based on the total of the density values of the specific pixel and the peripheral pixel and the density value of the pixel of interest. 4. The specific pixel is a pixel immediately before the pixel of interest on a raster scanning line.
The image coding method according to any one of items 1 to 3. 5. The peripheral pixel includes a pixel on a scanning line immediately preceding the pixel of interest.
The image coding method according to any one of items. 6. An image coding apparatus for coding image data, wherein the extracting means extracts a target pixel of image data, a pixel in the vicinity of the target pixel, and peripheral pixels of the target pixel as a specific pixel and a reference pixel, respectively. Based on the density difference between the target pixel and the specific pixel and the difference, a calculating unit that calculates a difference between the density values of the peripheral pixel and the specific pixel, a comparing unit that compares the difference with a predetermined value, Markov model encoding means for encoding the density difference Markov model encoding means, predictive encoding means for predictively encoding the pixel of interest based on the density average of the specific pixel and the peripheral pixel and the density value of the pixel of interest, Selection means for encoding the pixel of interest by using one of the Markov model encoding means and the prediction encoding means according to the comparison result by the comparison means. An image encoding device characterized by the above. 7. When the density difference between the pixel of interest and the specific pixel is equal to or larger than a predetermined value, the density difference is Markov model coded, and the pixel of interest is predictively coded using the specific pixel as a prediction value. The image coding apparatus according to claim 6, wherein 8. An image coding apparatus for coding image data, wherein the extracting means extracts a target pixel of image data, a pixel in the vicinity of the target pixel, and peripheral pixels of the target pixel as a specific pixel and a reference pixel, respectively. A calculating means for obtaining a difference in density value between the peripheral pixel and the specific pixel; a comparing means for comparing the difference with a predetermined value; and the target pixel based on the target pixel, the specific pixel and the peripheral pixel A Markov model coding means for coding a Markov model, a predictive coding means for predictively coding the pixel of interest based on density values of the specific pixel, the peripheral pixel, and the density value of the pixel of interest, and the comparison Selection means for encoding the pixel of interest by using one of the Markov model encoding means and the predictive encoding means according to the comparison result by the means. An image encoding device characterized by: 9. The specific pixel is a pixel immediately before the pixel of interest on a raster scanning line.
The image coding device according to any one of items 1 to 8. 10. The pixel according to claim 6, wherein the peripheral pixel includes a pixel on a scanning line immediately before the pixel of interest.
The image encoding device according to any one of items 8.